1
1
The passive surveillance of ticks using companion animal electronic health records
2
J.S.P. Tulloch * 1, L. McGinley 1, F. Sánchez-Vizcaíno 1,2, J.M. Medlock 1,3,4, A.D. Radford 1,5
3
4
5
1 NIHR Health Protection Research Unit in Emerging and Zoonotic Infections, University of Liverpool,
UK
6
7
2 Institute of Infection and Global Health, The Farr Institute@HeRC, University of Liverpool,
Waterhouse Building (2nd Floor, Block F), 1-5 Brownlow Street, Liverpool, L69 3GL, UK
8
9
3 Medical Entomology Group, Emergency Response Department, Public Health England, Porton
Down, Salisbury, SP4 0JG, UK
10
4 NIHR Health Protection Research Unit in Health and the Environment, Porton Down, SP4 0JG, UK
11
12
5 Institute of Infection and Global Health, University of Liverpool, Leahurst Campus, Chester High
Road, Neston, S. Wirral, CH64 7TE, UK
13
* Corresponding author
14
15
Email addresses:
16
J.S.P. Tulloch
[email protected]
17
L. McGinley
[email protected]
18
F. Sanchez-Vizcaino
[email protected]
19
J.M. Medlock
[email protected]
20
A.D. Radford
[email protected]
21
22
PASSIVE SURVEILLANCE OF TICKS USING COMPANION ANIMAL EHRs
2
23
Summary
24
Ticks represent a large global reservoir of zoonotic disease. Current surveillance systems can be time
25
and labour intensive. We propose that the passive surveillance of companion animal electronic
26
health records (EHRs) could provide a novel methodology for describing temporal and spatial tick
27
activity. A total of 1,658,857 EHRs were collected over a two year period (31st March 2014 and 29th
28
May 2016) from companion animals attending a large sentinel network of 192 veterinary clinics
29
across Great Britain (The Small Animal Veterinary Surveillance Network - SAVSNET). In total, 2,180
30
EHRs were identified where a tick was recorded on an animal. The relative risk of dogs presenting
31
with a tick compared to cats was 0.73 (95% confidence intervals, 0.67-0.80). The highest number of
32
tick records were in the south central regions of England. The presence of ticks showed marked
33
seasonality with summer peaks, and a secondary smaller peak in autumn for cats; ticks were still
34
being found throughout most of Great Britain during the winter. This suggests that passive
35
surveillance of companion animal EHRs can describe tick activity temporally and spatially in a large
36
cohort of veterinary clinics across Great Britain. These results and methodology could help inform
37
veterinary and public health messages as well as increase awareness of ticks and tick-borne diseases
38
in the general population.
39
Key words: Ticks, surveillance, companion animals, electronic health records, Great Britain, one
40
health
41
42
43
44
PASSIVE SURVEILLANCE OF TICKS USING COMPANION ANIMAL EHRs
3
45
Introduction
46
Ticks are effective vectors of zoonotic pathogens, and tick-borne diseases (TBDs) can be severely
47
debilitating to both humans and companion animals, in some cases leading to death. Lyme disease is
48
the most common TBD in the Northern Hemisphere with, in Western Europe, an unweighted mean
49
for annual incidence rate of 56.3/100,000 persons per year [1]. Tick borne diseases can pose a large
50
burden on health services; a recent study of Lyme borreliosis inpatients in Germany estimated an
51
annual cost in excess of 30 million Euros [2]. Due to this, and increasing public concern, governments
52
and research organisations are trying to heighten and improve their understanding of risk models of
53
ticks [3]. The responses to this call have largely fallen into three categories: the active and passive
54
collection of ticks, the utilisation of digital applications, and the monitoring of electronic health
55
records.
56
In Great Britain, Public Health England (PHE) coordinates the Tick Surveillance Scheme (TSS) [4,5],
57
which relies on passive submission of ticks by members of the public as well as medical / veterinary
58
professionals. Between 2005 and 2016 the TSS received a total of 18,000 ticks, primarily found on
59
companion animals and humans (PHE unpublished). Similarly, ‘The Big Tick Project’, collected data
60
on ticks found on dogs in the United Kingdom (UK) [6,7]; in their most recent study, 6,555 ticks were
61
actively collected from dogs attending select veterinary clinics over a 16 week period from April to
62
July 2015 [6]. Since such systems collect the actual tick, they are able to both identify the ticks and
63
describe their spatial distributions. However, they are labour and time intensive, relying on large
64
amounts of public engagement and involvement, and in the case of the Big Tick Project, do not
65
provide continuous surveillance data.
66
A very different approach has been developed by the National Institute for Public Health and the
67
Environment (RIVM) in The Netherlands. The Tekenradar website and digital app allows members of
68
the public to record when they have been bitten by a tick (and send it to RIVM), or develop an
PASSIVE SURVEILLANCE OF TICKS USING COMPANION ANIMAL EHRs
4
69
erythema migrans rash which is pathognomonic for Lyme disease [8,9]. This enables ‘live’ reporting
70
of tick bites and identifies areas of tick bite and Lyme disease risk. Due to its presence on multiple
71
digital platforms it facilitates easy promotion for public health messaging. It has also been promoted
72
as a resource for researchers of ticks and TBDs [9]. However, the success of such a system is largely
73
reliant on the accurate diagnosis and identification of ticks, tick bites and erythema migrans by
74
members of the public, rather than qualified health care professionals.
75
In Switzerland, the government has set up a voluntary surveillance system of 150 primary care
76
physicians called Sentinella [10], recording 1,644 cases of tick bites from 2008 to the end of 2011.
77
Rather than collecting and submitting ticks for further analyses, this system relies on the accurate
78
diagnosis and recording of tick bites by medical practitioners within their patients’ electronic health
79
records (EHRs), without the actual visualisation or collection of the tick. In a similar way, PHE use
80
routine passive syndromic surveillance based on a predetermined list of clinical codes to monitor the
81
incidence of arthropod bites in near real-time across various clinical settings including general
82
practitioner consultations, emergency department attendance and telephone helplines [11].
83
However constraints of the clinical diagnostic codes being used mean tick bites cannot be analysed
84
separately from those of other arthropods.
85
While each of these systems contribute to different aspects of tick surveillance, none of them
86
currently provide a surveillance system that is low cost and in sufficient temporal and spatial
87
resolution to quickly and efficiently provide large sets of data about generic tick activity.
88
According to the most recent estimates, there are 11.6 million dogs and 10.1 million cats kept as
89
pets in the UK, with 30% and 23% of households owning a dog and cat respectively [12]. These
90
species have the potential for greater exposure to tick habitats than humans, and often without
91
measures to prevent tick contact. It has been shown that dogs that are regularly walked are likely to
92
acquire ticks, and it is well established that dogs have the potential to act as sentinels for ticks and
PASSIVE SURVEILLANCE OF TICKS USING COMPANION ANIMAL EHRs
5
93
TBDs [6,13–16]. Due to owner concern, companion animals with ticks are often presented to
94
veterinary clinics, with the veterinary practitioner frequently recording the presence of ticks within
95
an individual animal’s electronic health record (EHR) [17]. The aim of this paper is to explore the
96
feasibility of using such EHRs from a large sentinel network of veterinary clinics as the basis of a
97
novel surveillance system to provide efficient temporal and spatial estimates of tick activity risk in
98
Great Britain that complement existing tick surveillance schemes.
99
100
Methods
101
EHRs were collected through the Small Animal Veterinary Surveillance Network (SAVSNET) from
102
volunteer veterinary clinics using a compatible practice management system; currently Teleos ™ and
103
RoboVet™. This study uses over two years of data gathered from 192 veterinary clinics across the UK
104
between 31st March 2014 and 29th May 2016 (Figure 1). Each EHR was collected at the end of a
105
veterinary consultation in real-time and included the following data; date of the consultation,
106
postcode of the owner, species of the animal in the consultation and the clinical narrative, which
107
would have been written by the consulting veterinary surgeon or nurse. Whilst data on
108
ectoparasiticide treatments were collected, they were not included in further analysis; many are
109
active against multiple arthropods and routine prophylactic prescription by veterinarians results in a
110
low specificity for tick infestation.
111
Initially, a simple free text analysis approach was developed to identify EHRs containing the word
112
‘tick’ in the clinical narrative field, whilst excluding records where only the terms ‘tickl’, ‘ticki’ and
113
‘sticki’ were used. To increase the specificity of such an approach, the resulting EHRs were
114
subsequently read by two domain experts (JT and LM) who verified the reference to ticks based on a
115
strict case definition. A tick was only deemed present in a consultation if, within the associated EHR,
PASSIVE SURVEILLANCE OF TICKS USING COMPANION ANIMAL EHRs
6
116
‘a veterinary surgeon or nurse confirmed visual sighting or removal of a tick within the consultation.’
117
This case definition was used to avoid any misidentification errors of ticks by owners, or any
118
historical identification of ticks being included in the current analysis (e.g. tick removed last week).
119
To verify concordant interpretation of the case definition, the two domain experts manually
120
classified a random sample of the EHRs. The amount of agreement (i.e. inter-rate reliability) was
121
measured by Cohen’s kappa coefficient (R package ‘psych’) [18]. Each EHR where an agreement was
122
not achieved was re-examined by both domain experts to develop a more consistent interpretation
123
of the case definition. This process was repeated until an ‘almost perfect’ level of agreement was
124
achieved, at which point the remaining EHRs were randomly divided between the two domain
125
experts and categorised [19].
126
Results of these analyses were used to calculate the number of consultations where a tick was
127
recorded in the EHR per 10,000 consultations. We assume this to be a proxy for activity of ticks and
128
for brevity, refer to this measure as “tick activity”. Relative risks were calculated between the two
129
predominant host species (i.e. dogs and cats) with statistical significance (P<0.05) measured by a Chi-
130
squared test.
131
Time-series plots (based on the time of the consultation as recorded in the EHR) were used to
132
identify temporal trends in tick activity and to compare temporal trends by host species. The
133
temporal pattern of tick activity was smoothed using a nonparametric method, the LOESS (locally
134
weighted regression) technique (R package ‘ggplot2’) [20,21]. Outliers were identified as those
135
outside the smoothed line’s 95% confidence intervals. All proportions and 95% confidence intervals
136
(CIs) were calculated using robust standard errors to account for intragroup correlation within
137
veterinary clinics. Statistical analyses were carried out using R language (version 3.2.0) (R Core Team
138
2015).
PASSIVE SURVEILLANCE OF TICKS USING COMPANION ANIMAL EHRs
7
139
Maps were used to describe the spatial distribution of tick activity during each season. We defined
140
season as winter (December-February), spring (March-May), summer (June-August) and autumn
141
(September-November). The spatial distribution of tick activity was stratified by owner’s given
142
address. SAVSNET receives full owner postcode for each EHR, which locates each address to one of
143
1.75 million locations [22]. However at such resolution it is possible to identify some individual
144
properties, particularly in rural areas. Therefore, postcode area (first half of postcode; n=124) was
145
used to maintain owner confidentiality when presenting the results. When displaying the data, we
146
took a cautious approach and excluded areas with less than 200 EHRs in each season, as they were
147
less likely to be representative. A map was constructed displaying all EHRs, aggregated by owner’s
148
postcode area, to show the underlying population distribution (Figure 1). The data was depicted
149
using QGIS version 2.8.2-Wien.
150
151
Results
152
In total 1,658,857 EHRs were collected during the study period, consisting of 70.5% dogs and 26.4%
153
cats. Of these, 10,155 (0.61%) had a clinical narrative containing the word ‘tick’. The two domain
154
experts first independently read and applied the case definition to 365 randomly selected EHRs from
155
these 10,155. After adjusting by the amount of agreement which would be expected by chance, a
156
‘substantial agreement’ (K= 0.7; 95% CI: 0.63-0.78) with 305 EHRs agreed was achieved. After
157
reappraising this first data set, the exercise was repeated on a new random sample of 365 EHRs, this
158
time achieving an ‘almost perfect’ agreement (K= 0.82; 95%CI: 0.77-0.88) with 332 EHRs agreed; the
159
remaining EHRs were therefore categorised independently by the two authors.
160
In total, 2,180 EHRs were confirmed as having a tick present, equating to 0.13% of the total
161
1,658,857, and 21.5% of the 10,155 automatically identified EHRs. Of these 2,180 EHRs, 1,421 were
PASSIVE SURVEILLANCE OF TICKS USING COMPANION ANIMAL EHRs
8
162
from dogs (65.2%), 728 from cats (33.4%), and 17 from other species (which only included ferrets,
163
rabbits and guinea pigs; 0.8%), with the remaining 14 EHRs lacking an identifiable species label
164
(0.6%). The relative risk of a dog being recorded as presenting with a tick compared to that for a cat
165
was 0.73 (95% CI: 0.67-0.80, p<0.005). The main reasons for EHRs being identified by the free text
166
analysis but failing to meet the case definition included; misidentification of ticks by owners (e.g.
167
skin tags, nipples, tumours), ticks observed by owners before the consultation and not confirmed by
168
a veterinary surgeon or nurse within the consultation, and discussions held in the consultation about
169
ticks and TBDs without a tick being present. Only five of the 2,180 (0.2%) EHRs identified as relating
170
to ticks included information at genus and species level; two referring to Ixodes spp, one to I.ricinus,
171
one to Dermacentor spp and one EHR referring to both Dermacentor and Rhipicephalus spp.
172
The mean weekly rate of tick reporting in this population over the entire study period was 15.3 tick
173
based EHRs per 10,000 EHRs. The temporal pattern was similar in both calendar years, with peak tick
174
activity between May and July each year, and highest levels recorded in mid-June (Figure 2a).
175
Minimum tick activity was between December and February, with the lowest activity in January in
176
both calendar years. The temporal pattern of tick activity in dogs was similar to the overall
177
population, with peak activity in June (maximum of 65.5 tick based EHRs per 10,000 EHRs over a
178
single week) and lowest levels between December and February (Figure 2b). In contrast, cats
179
seemed to have an earlier peak in weekly tick activity in May (with a maximum of 87.2 tick based
180
EHRs per 10,000 EHRs), with a secondary smaller peak in the autumn, and their lowest levels in
181
February (Figure 2b). In the winter of 2015-2016, ticks were still recorded in every week on cats,
182
whilst for two separate weeks none were recorded on dogs. The mean weekly rate of tick activity
183
was lower for dogs (14.8 tick based EHRs per 10,000 EHRs) than for cats (18.3 tick based EHRs per
184
10,000 EHRs).
185
There was considerable variation in the spatial distribution of tick activity across each postcode area
186
in Great Britain (Figure 3). The ten postcode areas with highest tick activity across all seasons were
PASSIVE SURVEILLANCE OF TICKS USING COMPANION ANIMAL EHRs
9
187
(in descending numerical order); Bournemouth, Hemel Hempstead, Southampton, Falkirk, Salisbury,
188
Guildford, Croydon, Llandudno, Reading and Lancaster. Of these, Southampton, Bournemouth,
189
Guildford, Reading, Llandudno and Lancaster peaked in spring; the remainder peaking in summer.
190
No postcode areas had their peak activity in autumn or winter. The areas with no tick activity across
191
all seasons were Hereford, Oldham and Wolverhampton.
192
193
Discussion
194
Ticks are an important vector of disease but continuous surveillance has proved challenging. Here
195
we show how text mining of companion animal EHRs from a large sentinel population of veterinary
196
clinics may provide a novel form of passive surveillance to describe temporal and spatial trends in
197
tick activity across Great Britain. This could inform more targeted public health messaging to the risk
198
of veterinary disease and human health risks associated with ticks.
199
Our data showed a seasonality to tick activity consistent with previous reports with peaks of activity
200
at the start of the summer and minimal activity during winter [4,23–25]. Although we are unable to
201
identify the life cycle stage of the ticks referred to within the EHRs, we assume this seasonality
202
largely reflects that of adult ticks, and to a lesser extent nymphs, of Ixodes ricinus, as these are the
203
most common ticks found on companion animals in Great Britain [4,6,7,24,26]. This tick is
204
susceptible to desiccation and its host seeking behaviour (questing) is greatly influenced by changes
205
in temperature and humidity [27]. It therefore has an annual variation of peak activity, with levels
206
rising in early spring and peaking between April and July, with low levels in winter [24], although this
207
may show regional variation [28]. The fact that the seasonal profile we observed in dogs was similar
208
to that in the total population, is a reflection of the demographic predominance of dogs in our data
209
and in other such veterinary visiting populations [29,30].
PASSIVE SURVEILLANCE OF TICKS USING COMPANION ANIMAL EHRs
10
210
Interestingly, tick activity on cats showed marked differences to that seen in dogs, raising several
211
important questions relating both to ectoparasite biology, as well as owner and veterinary surgeon
212
behaviour. Our study is the first to suggest that cats are more likely to present to veterinary clinics
213
with ticks than dogs. Previous studies based on tick submissions either excluded cats [6,7] or lacked
214
suitable population denominator data to calculate a relative risk [4]. Whether this represents a
215
genuine increased risk of ticks on cats, or that ticks on cats are more likely to be observed by owners
216
and presented to the veterinary surgery, or whether veterinary surgeons are more likely to record
217
ticks on cats in their EHRs, remains to be determined. As well as this overall increased risk, cats
218
continued to present with ticks during the winter of 2015-2016, in contrast to dogs where there
219
were short periods where ticks were not identified in this population. During these months, dog
220
owners may be less inclined to take their dogs for exercise where they may be exposed to ticks due
221
to shortened day length, cooler temperatures and higher rainfall. However, domestic cats in the UK
222
may remain susceptible to ticks, albeit at lower levels, due to their ability to explore outside habitats
223
at their own free will due to the common use of cat flaps. The indoor or outdoor nature of a cat is
224
not explicitly recorded within the SAVSNET population, unless it has been recorded within the EHR.
225
This potential risk factor was therefore not studied.
226
Ticks on cats also showed a different temporal pattern of tick activity with an earlier main peak in
227
the spring and some evidence for a second smaller peak in the autumn. The precise reason for this
228
apparent difference remains unknown but may relate to differences in host susceptibility to
229
different tick species. Cats are significantly more likely to carry Ixodes hexagonus than I. ricinus [26]
230
and I. hexagonus is more frequently found on cats than dogs [4,5,26]. The activity of I. hexagonus is
231
closely linked to the density and behaviour of its primary host, the European hedgehog (Erinaceus
232
europaeus) [31]. I. hexagonus is more prevalent earlier in the year than I. ricinus [31,32], coinciding
233
with the emergence of hedgehogs from hibernation [33], and possibly explaining the earlier peak of
234
tick activity we identified in cats. The second autumnal peak could represent interaction between
PASSIVE SURVEILLANCE OF TICKS USING COMPANION ANIMAL EHRs
11
235
cats and hedgehogs at a time when hedgehogs are preparing for hibernation and juveniles are
236
gaining independence, leading to greater hedgehog numbers being seen [33], all at a time when I.
237
hexagonus is also at great abundance on the hedgehogs themselves [31,32].
238
The spatial distribution we describe generally mirrors previous work on tick distributions in Great
239
Britain [4–7,23]. Comparing it to the most recent study published using data from a shorter, but
240
overlapping time period (16 weeks between April – July 2015), both studies identified the highest
241
levels of tick activity in southern postcode areas of England, with high levels also in the south of
242
Scotland [6]. However, in contrast, we see higher levels of activity in north and mid-Wales, and
243
north-west England, and less clear areas of high activity in north Norfolk and the north-east of
244
England. These observations likely reflect differences in methodology used by the two projects
245
including veterinary clinic recruitment, the period of sampling, and potentially tick distribution [6].
246
Collection of continuous surveillance data over two years across GB has allowed us to begin to
247
describe a complex mosaic of tick activity across the country in different seasons. However, broad
248
trends can be identified. In winter, low levels of tick activity remain throughout England and Wales,
249
challenging the belief of some vets, who recorded in their EHRs that ticks pose no risk in winter
250
(unpublished observations). The results also showed that the timing of peak activity varied by
251
postcode area, with the majority of areas peaking in the spring, the remainder peaking in the
252
summer. The data set described here represents a rich research tool in which to explore the varied
253
impact of climate, and other environmental and ecological factors, on tick activity.
254
To maintain a high specificity, we applied a very restrictive case definition, only including ticks that
255
were seen by a veterinary surgeon or nurse and recorded during the consultation. Therefore, it is
256
clear that not all ticks on cats and dogs will be included in our study. Many ticks on companion
257
animals will not present to the veterinary practice either because the owner is not concerned, or
258
removed the tick themselves, or the ticks were not noticed. Equally ticks on animals in a veterinary
PASSIVE SURVEILLANCE OF TICKS USING COMPANION ANIMAL EHRs
12
259
consultation may not be noticed, or not recorded, especially where they are incidental findings in
260
relation to what may be a more serious clinical need. Indeed, where dogs had a bespoke thorough
261
clinical examination as part of a research study to identify tick carriage, reported tick prevalence was
262
much higher (30%) [6]. This study was however carried out during peak tick activity (April-July) and
263
as the authors stated, practitioners participating in the study may have been more likely to sample
264
animals with observed ticks on them. Although it is clear that the values we report are therefore an
265
underestimate of overall tick activity on companion animals, we feel confident that they can
266
describe relevant levels of relative risk. We must also acknowledge that health scares and media
267
coverage could influence owner behaviour and veterinary recording behaviour. This has been
268
previously discussed in relation to the Babesia canis outbreak seen in early 2016 [17]. However, in
269
this particular case, this outbreak did not appear to influence the overall temporal trends of our data
270
(data not presented).
271
Arrival of exotic ticks has been of great concern to both the veterinary and medical professions as
272
they have the potential to carry pathogens not currently transmitted in the UK [34–37]. This has
273
driven a need for species level surveillance of ticks such as provided by PHE [5] and the Big Tick
274
project [6]. Within our data, only five EHRs included information at the genus and species level; two
275
referring to Ixodes spp, one to I. ricinus, one to Dermacentor spp and one EHR referring to both
276
Dermacentor and Rhipicephalus spp. Although these numbers are clearly low and in the absence of
277
microscopic confirmation need to be treated with some caution, they still raise important questions.
278
Whilst a few foci of Dermacentor are known to exist in the UK [5], Rhipicephalus sanguineus has only
279
been reported in dogs that have travelled in the rest of Europe [6,37], such that reference to
280
Rhipicephalus spp in even one EHR could be significant. The infrequent mention of tick species likely
281
reflect time constraints of a short consultation, the challenge of identification, especially if the tick is
282
engorged, and veterinary surgeons deeming it clinically irrelevant. In the future, the reference to
283
rare and exotic tick species identified by EHR surveillance, could be followed up by submission of the
PASSIVE SURVEILLANCE OF TICKS USING COMPANION ANIMAL EHRs
13
284
tick to relevant health authorities with tick identification capabilities; such as PHE. Surveillance
285
systems based on EHRs would be improved if veterinary surgeons were encouraged to record within
286
the EHR any recent travel history and information about tick species where they are confident to do
287
so.
288
The current limitations of this study are inherent to its methodology. Since recruitment of practices
289
is not random, there may be selection bias in our results, meaning generalisability to the entire UK
290
population of veterinary visiting dogs and cats is not possible. In addition, population statistics for
291
companion animals in the UK are generally poor or unavailable, such that our results cannot be
292
described by incidence; this may change as compulsory microchipping of dogs has recently come
293
into legislation [38]. Some postcode areas have relatively small amounts of data and were excluded
294
from our analyses. However, the fact that 56% (70 of 124) of postcode areas contributed more than
295
5,000 EHRs during the study period, and that 42% (52 of 124) of areas contributed more than 10,000
296
EHRs suggests that we already have good data coverage for large parts of Great Britain. As SAVSNET
297
continues to expand through clinic recruitment, we believe that the spatial distribution of clinics and
298
the number of EHRs collected will become more homogenous. Our data will always underestimate
299
true tick activity on companion animals, and veterinary surgeons or nurses rarely record the tick
300
species in EHR. In addition, like other studies that define a tick’s location by the pet owner’s
301
postcode [6], our results should be seen as a proxy for tick activity at a given geographical area,
302
rather than the location where the animal necessarily acquired the tick.
303
Despite these current limitations, we believe this form of surveillance offers some real benefits.
304
Using EHRs is very passive in nature, as once a veterinary clinic has been enrolled, no changes in
305
clinician behaviour need occur for the data to be captured. This data is collected in real-time, with
306
the only rate-limiting step currently being the time taken to verify the strict case definition. Research
307
is currently being undertaken to enhance the specificity of our free text analysis approach by
308
utilising natural language processing, this will improve the level of automation and reduce the
PASSIVE SURVEILLANCE OF TICKS USING COMPANION ANIMAL EHRs
14
309
number of cases to be verified. Compared to systems that rely on the general public identifying a
310
tick, ticks recorded in EHRs are identified by a qualified health care professional [9]. There is also
311
minimal labour required, except the upkeep of a system to collect the EHRs on which it relies. As our
312
results are similar to previous surveillance and field work performed in the UK, we believe that this
313
method provides a novel and complementary approach for tick surveillance that could be adopted
314
by other countries where mature pet animal EHRs exist. As more clinics are recruited the
315
representativeness of such systems can be improved. Linking data through postcode to other data
316
sources, such as habitat type and localised meteorological data, will provide new opportunities to
317
understand the effect of climate change and land use changes on the distribution and activity levels
318
of ticks [34,35,39–41].
319
In summary, this study shows how the passive real-time collection of companion animal EHRs can
320
provide efficient, accurate and novel data on tick activity in a large national sentinel population of
321
companion animals. We highlight for the first time, temporal differences of tick exposure between
322
domesticated cats and dogs. As the availability of EHRs increases, such methodology can provide a
323
comprehensive temporal and spatial understanding of tick activity, and in combination with other
324
systems already in place, has the potential to further inform tick and TBD risk models, aiding a ‘One-
325
Health’ approach for public health messaging and tick control.
326
327
List of Abbreviations
328
TBD
-
Tick-borne disease
329
EHR
-
Electronic Health Record
330
SAVSNET
-
Small Animal Veterinary Surveillance Network
PASSIVE SURVEILLANCE OF TICKS USING COMPANION ANIMAL EHRs
15
331
PHE
-
Public Health England
332
TSS
-
Tick Surveillance Scheme
333
UK
-
United Kingdom
334
RIVM
-
National Institute for Public Health and the Environment
335
CI
-
Confidence Intervals
336
337
Declarations
338
Ethics approval and consent to participate
339
Ethics approval for this project came from the University of Liverpool research committee
340
(RETH00964). Consent to participate is recognised via an opt out process available to all companion
341
animal owners at veterinary clinics participating in SAVSNET.
342
Acknowledgements
343
Dr Jenny Newman and David Singleton for their continued work and development of SAVSNET.
344
Sammuel Pettinger, a Nuffield scholar, who did some preliminary work with SAVSNET data on ticks.
345
Stephanie Begemann for her translation skills. We would like to thank all SAVSNET member clinics,
346
without whom this research could not have occurred.
347
Financial Support
PASSIVE SURVEILLANCE OF TICKS USING COMPANION ANIMAL EHRs
16
348
JT, LM and FSV are fully supported, and AR and JMM partly supported by the National Institute for
349
Health Research Health Protection Research Unit (HPRU) in Emerging and Zoonotic Infections at
350
University of Liverpool in partnership with Public Health England (PHE), in collaboration with
351
Liverpool School of Tropical Medicine. JT, LM, FSV and AR are based at the University of Liverpool.
352
The views expressed are those of the authors and not necessarily those of the NHS, the NIHR, the
353
Department of Health or Public Health England. JMM is also partly funded by the HPRU in
354
Environmental Change and Health. JMM is based at Public Health England. Data costs were met by
355
SAVSNET. SAVSNET is funded by the British Small Animal Veterinary Association (BSAVA) with
356
additional support by The Animal Welfare Foundation (AWF).
357
Availability of data and materials
358
The datasets generated during and/or analysed during the current study are not publicly available
359
due issues of companion animal owner confidentiality, but are available on request from the
360
SAVSNET Data Access and Publication Panel ([email protected]) for researchers who meet the
361
criteria for access to confidential data.
362
Author’s contributions
363
FSV extracted the SAVSNET data set and provided epidemiological and statistical support. JT and LM
364
read all the consults. JT performed the data analysis. AR designed and managed the study, and
365
developed the SAVSNET database. JMM for interpretation of the data. All authors reviewed and
366
approved the final manuscript.
367
Conflict of Interest
368
None
PASSIVE SURVEILLANCE OF TICKS USING COMPANION ANIMAL EHRs
17
369
Consent for publication
370
Provided by the National Institute for Health Research and the SAVSNET Data Access and Publication
371
Panel (DAPP)
372
Disclaimer/Declarations
373
374
375
376
377
378
379
The research was funded by the National Institute for Health Research Health Protection Research
Unit (NIHR HPRU) in Emerging and Zoonotic Infections at the University of Liverpool in partnership
with Public Health England (PHE) and Liverpool School of Tropical Medicine (LSTM). The views
expressed are those of the author(s) and not necessarily those of the NHS, the NIHR, the
Department of Health or Public Health England. JMM was also partly funded by the NIHR HPRU in
Environment and Health led by the London School of Hygiene and Tropical Medicine in partnership
with Public Health England, the University of Exeter and the Met Office.
380
381
References
382
1
383
384
Public Health 2016; :1–8.
2
385
386
3
391
Norman RA, Worton AJ, Gilbert L. Past and future perspectives on mathematical models of
tick-borne pathogens. Parasitology 2015; :1–10.
4
389
390
Lohr B et al. Epidemiology and cost of hospital care for Lyme borreliosis in Germany: Lessons
from a health care utilization database analysis. Ticks and Tick-borne Diseases 2015; 6:56–62.
387
388
Sykes RA, Makiello P. An estimate of Lyme borreliosis incidence in western Europe. Journal of
Jameson LJ, Medlock JM. Tick surveillance in Great Britain. Vector borne and zoonotic
diseases (Larchmont, NY) 2011; 11:403–12.
5
PHE - Tick Surveillance Scheme. 2016.https://www.gov.uk/guidance/tick-surveillance-scheme
(accessed 9 Sep2016).
PASSIVE SURVEILLANCE OF TICKS USING COMPANION ANIMAL EHRs
18
392
6
393
394
Abdullah S et al. Ticks infesting domestic dogs in the UK: a large-scale surveillance
programme. Parasites & Vectors 2016; 9:391.
7
395
Smith FD et al. Prevalence, distribution and risk associated with tick infestation of dogs in
Great Britain. Medical and Veterinary Entomology 2011; 25:377–384.
396
8
Tekenradar. https://www.tekenradar.nl/ (accessed 25 Jul2016).
397
9
Harms MG et al. Onderzoek in het kort - Tekenradar . nl, een webplatform over tekenbeten
398
399
en de ziekte van Lyme. Infectieziekten Bulletin 2014; 7:204–206.
10
400
401
Altpeter E et al. Tick related diseases in Switzerland, 2008 to 2011. Swiss Medical Weekly
2013; 143:1–13.
11
Newitt S et al. The use of syndromic surveillance to monitor the incidence of arthropod bites
402
requiring healthcare in England , 2000 – 2013 : a retrospective ecological study. Epidemiology
403
and infection 2016; 144:2251–2259.
404
12
405
406
household ownership. The Veterinary Record 2015; 177:259.
13
407
408
411
Goossens HAT, Van den Bogaard AE, Nohlmans MKE. Dogs as sentinels for human Lyme
borreliosis in The Netherlands. Journal of Clinical Microbiology 2001; 39:844–848.
14
409
410
Murray JK et al. Assessing changes in the UK pet cat and dog populations: numbers and
Smith FD et al. Estimating Lyme disease risk using pet dogs as sentinels. Comparative
Immunology, Microbiology and Infectious Diseases 2012; 35:163–167.
15
Herrmann JA; Dahm NM, Ruiz MO BW. Temporal and spatial distribution of tick-borne
disease cases among humans and canines in Illinois (2000-2009). Environmental Health
PASSIVE SURVEILLANCE OF TICKS USING COMPANION ANIMAL EHRs
19
412
413
Insights 2009; 8:8–10.
16
414
415
Jennett AL, Smith FD, Wall R. Tick infestation risk for dogs in a peri-urban park. Parasites &
Vectors 2013; 6:358.
17
Sánchez-Vizcaíno F et al. Canine babesiosis and tick activity monitored using companion
416
animal electronic health records in the UK. The Veterinary Record 2016;
417
:doi:10.1136/vr.103908.
418
18
419
420
Measurement 1960; 20:37–46.
19
421
422
20
Cleveland WS. Robust locally weighted regression and smoothing scatterplots. Journal of the
American Statistical Association 1979; 74:829–836.
21
425
426
Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics
1977; 33:159–174.
423
424
Cohen J. A Coefficient of agreement for nominal scales. Educational and Psychological
Jacoby WG. Loess: A nonparametric, graphical tool for depicting relationships between
variables. Electoral Studies 2000; 19:577–613.
22
Office for National Statistics - Postal Geography.
427
2016.http://webarchive.nationalarchives.gov.uk/20160105160709/http://www.ons.gov.uk/o
428
ns/guide-method/geography/beginner-s-guide/postal/index.html (accessed 9 Sep2016).
429
23
430
431
Pietzsch ME et al. Distribution of Ixodes ricinus in the British Isles: Investigation of historical
records. Medical and Veterinary Entomology 2005; 19:306–314.
24
Randolph SE et al. An empirical quantitative framework for the seasonal population dynamics
PASSIVE SURVEILLANCE OF TICKS USING COMPANION ANIMAL EHRs
20
432
433
of the tick Ixodes ricinus. International Journal for Parasitology 2002; 32:979–989.
25
Dobson ADM, Taylor JL, Randolph SE. Tick (Ixodes ricinus) abundance and seasonality at
434
recreational sites in the UK: Hazards in relation to fine-scale habitat types revealed by
435
complementary sampling methods. Ticks and Tick-borne Diseases 2011; 2:67–74.
436
26
437
438
and Ireland. Medical and Veterinary Entomology 2000; 14:332–338.
27
439
440
Ogden NH et al. The ixodid tick species attaching to domestic dogs and cats in Great Britain
Qviller L et al. Temporal pattern of questing tick Ixodes ricinus density at differing elevations
in the coastal region of western Norway. Parasites & vectors 2014; 7:179.
28
Gilbert L, Aungier J, Tomkins JL. Climate of origin affects tick (Ixodes ricinus) host-seeking
441
behavior in response to temperature: Implications for resilience to climate change? Ecology
442
and Evolution 2014; 4:1186–1198.
443
29
444
445
Robinson NJ et al. Investigating common clinical presentations in first opinion small animal
consultations using direct observation. The Veterinary Record 2015; 176:463.
30
Radford AD et al. Antibacterial prescribing patterns in small animal veterinary practice
446
identified via SAVSNET: the small animal veterinary surveillance network. The Veterinary
447
record 2011; 169:310.
448
31
Pfäffle M et al. Comparative population dynamics of a generalist (Ixodes ricinus) and specialist
449
tick (I. hexagonus) species from European hedgehogs. Experimental and Applied Acarology
450
2011; 54:151–164.
451
452
32
Dziemian S et al. Ectoparasite loads in sympatric urban populations of the northern whitebreasted and the European hedgehog. Parasitology Research 2015; 114:2317–2323.
PASSIVE SURVEILLANCE OF TICKS USING COMPANION ANIMAL EHRs
21
453
33
454
455
Thesis, University College Cork 2011.
34
456
457
35
Medlock JM, Leach SA. Effect of climate change on vector-borne disease risk in the UK. The
Lancet Infectious Diseases 2015; 15:721–730.
36
460
461
Gray JS et al. Effects of climate change on ticks and tick-borne diseases in Europe.
Interdisciplinary Perspectives on Infectious Diseases 2009; 2009. doi:10.1155/2009/593232
458
459
Haigh AJ. The ecology of the European hedgehog (Erinaceus europaeus) in rural Ireland. PhD
Phipps LP et al. Babesia canis detected in dogs and associated ticks from Essex. Veterinary
Record 2016; 178:243.3-244.
37
462
Hansford KM et al. Importation of R sanguineus into the UK via dogs: tickborne diseases. The
Veterinary record 2014; 175:385–6.
463
38
The Microchipping of Dogs (England) Regulations 2015; 2015.
464
39
Lindgren E, Talleklint L, Polfeldt T. Impact of climatic change on the northern latitude limit
465
and population density of the disease-transmitting European tick Ixodes ricinus.
466
Environmental Health Perspectives 2000; 108:119–123.
467
40
468
469
470
Scharlemann JPW et al. Trends in ixodid tick abundance and distribution in Great Britain.
Medical and Veterinary Entomology 2008; 22:238–247.
41
Medlock JM et al. Driving forces for changes in geographical distribution of Ixodes ricinus ticks
in Europe. Parasites & Vectors 2013; 6:1–11.
471
PASSIVE SURVEILLANCE OF TICKS USING COMPANION ANIMAL EHRs
22
472
473
474
475
476
477
Figures and Legends
478
Figure 1 – The distribution of participating SAVSNET (The small animal veterinary surveillance
479
network) veterinary clinics (red dots) within Great Britain, and the total number of electronic health
480
records collected between April 2014 and May 2016 by owners’ postcode area.
481
Figure 2a – Time series plot showing the weekly number of tick based electronic health records
482
(EHRs) per 10,000 EHRs between April 2014 and May 2016, in Great Britain
483
Figure 2b – Time series plot showing the weekly number of tick based electronic health records
484
(EHRs) per 10,000 EHRs in dogs and cats between April 2014 and May 2016, in Great Britain
485
Figure 3 – Geographical distribution of tick based electronic health records (EHRs) per 10,000 EHRs in
486
Great Britain, aggregated by owners’ postcode area for each season between April 2014 and May
487
2016. The dotted postcode areas represent areas with less than 200 EHRs in total during the relevant
488
time period.
PASSIVE SURVEILLANCE OF TICKS USING COMPANION ANIMAL EHRs